TEE based Agreement and Semantic Smoothing Combats Disinformation
The rapid pace of social media has exacerbated the challenge of countering disinformation, as both malicious actors and well-meaning individuals often misrepresent facts. The 2024 US Election Campaigns illustrate this issue, with both sides manipulating narratives through counter-factuals and omissions. Another pertinent example involves a recent altercation between Manchester Police and individuals at Manchester Airport. This event was swiftly exploited by various social media factions, including human rights activists, police watchdogs, secure border advocates, Muslim communities, and right-wing groups.
The Manchester Airport Incident
The incident involved several key components:
- A public disturbance by a minority group
- Failure to comply with police officers
- An altercation resulting in injuries to officers, including a female officer with a broken nose
- Allegations of police over-response and excessive force
- Rapid spread of partial facts through social media
The complexity of this scenario made it a prime target for those looking to provoke emotive responses from a global audience. Traditional responses, such as investigations and formal reports, are often delayed and subject to bias from controlling groups. For instance, a miscaptioned photo showed a female officer with a severe nose injury, implying it was from the incident and justifying the police response. However, this photo was unrelated to the event.
The Challenge of Disinformation
Disinformation can be driven by both unethical motives (to sway public opinion) and ethical ones (to counter existing narratives). Regardless of the motivation, the core issues remain:
- How do we stop misleading or false narratives?
- How can we ensure that people don’t feel compelled to act as “information arbiters”?
The Role of Trusted Execution Environments (TEEs)
Trusted Execution Environments (TEEs) offer a revolutionary approach to address these issues within the web3 ecosystem. By leveraging a DeCC (Decentralized Confidential Compute) compatible network, TEEs can facilitate both fact-convergence and agreement among different observers.
TEN Network: Enhancing Data Security and Integrity
The TEN network provides a decentralized, scalable platform based on TEE technology. It is fully encrypted, EVM compatible L2 protocol. This setup ensures the secure and private processing of data, which is crucial technical component for effectively combating disinformation at scale. With TEN’s network design we can mitigate the problem by reducing it to two levels of abstract consensus:
Level 1: Consensus on Probable Truth (Agreement)
Given numerous subjective perceptions, TEEs on the TEN network help establish a threshold of probable truth or certainty based on axioms or prior truths. This process is akin to pricing algorithms/markets, where the goal is to determine or guarantee a fair and accurate price. TEEs enhance this by enabling the use of vast amounts of data without leaking or identifying sensitive information.
Level 2: Consensus on Shared Meaning (Semantic Smoothing)
This level addresses the challenge of ensuring that observers interpret facts similarly. In a blockchain analogy, imagine two nodes (people) who speak the same language and use the same data representations. The challenge arises when these nodes (people) have different languages or interpretations. TEEs on the TEN network mitigate this by using private agents to transform and align user models securely*, creating a shared understanding.
*For more detail and information on how these transformations may work read up on mNFTs.
Practical Application of TEEs on the TEN Network
TEEs on the TEN network enable techniques that decouple ambiguity and misrepresentation of data. For example, in the case of the Manchester Airport incident, an over-exaggerated misrepresentation might have been used to evoke a more factual interpretation amid external noise. TEEs can adjust the core reality to match individual sensitivities, creating a consensus on the relativity of the event for each observer.
Illustration 1: Consensus on Probable Truth
Example: Investigating a Car Accident
Imagine there’s a car accident at a busy intersection. Multiple witnesses see the accident, and several security cameras capture footage from different angles. News reports and social media posts quickly spread different versions of what happened.
- Data Collection:
— TEEs collect data from all sources: security camera footage, witness statements, social media posts, and news reports.
— Analogy: Think of a detective gathering all possible evidence from a crime scene, including videos, photos, and testimonies from people who saw the event. - Data Validation:
— The data is processed within the TEE to verify its authenticity and relevance. This includes checking the timestamps on videos, corroborating witness statements, and filtering out fake social media posts.
— Analogy: The detective checks the validity of each piece of evidence. For example, confirming that the videos are from the correct date and time and ensuring that the witnesses were actually present at the scene. - Consensus Mechanism:
— A consensus algorithm analyzes the validated data to determine the most likely sequence of events. It might weigh eyewitness accounts that match video footage more heavily than unsupported social media claims.
— Analogy: The detective pieces together the timeline of events by comparing all validated evidence, figuring out what most likely happened based on the most reliable sources. - Output:
— The TEE outputs a detailed report of the accident, including a timeline and the most likely scenario of events.
— Analogy: The detective writes a final report summarizing the accident, detailing the sequence of events backed by solid evidence, which is then shared with the parties involved (e.g., insurance companies, law enforcement).
Illustration 2: Consensus on Shared Meaning
Example: Communicating a Health Advisory
Imagine a new health advisory is issued about a sudden outbreak of a flu virus. Different communities might understand and react to this information differently based on their cultural and social backgrounds.
- User Model Collection:
— TEEs collect and maintain models of different user groups to understand how they interpret information. For example, understanding how elderly people, young adults, parents, and healthcare professionals might each react to health advisories.
— Analogy: Think of a translator who knows the cultural and language nuances of different groups and can adapt messages accordingly. - Data Transformation:
— The health advisory data is transformed within the TEE to align with the user models. This ensures the core message about the flu virus remains consistent but is presented in ways that are meaningful to different groups.
— Analogy: The translator adjusts the message based on the audience. For the elderly, the message might emphasize the importance of vaccination. For parents, it might focus on symptoms to watch for in children. - Private Agent Interaction:
— Private agents within the TEE facilitate interactions between the user models and the transformed data, ensuring a shared understanding. Each group receives a version of the advisory that resonates with them while maintaining the core facts.
— Analogy: The translator ensures that each group understands the critical health message in their own context, reducing misunderstandings and ensuring everyone takes the right actions. - Output:
— The transformed health advisory is outputted in different formats for different user groups, ensuring that everyone receives the same core information in a way they can best understand.
— Analogy: The translator delivers tailored messages to each group. Elderly individuals get a clear and straightforward advisory, young adults receive a version that addresses their concerns about social activities, and parents get information on how to protect their children.
Conclusion
Using these analogies, we can visualize how TEEs on the TEN network work to achieve consensus on probable truth and shared meaning. By securely collecting, validating, and transforming data, TEEs ensure that information is accurate. Further, TEEs enable augmenting to audience which reduces the desire to defensively-game against the truth. In tandem, these strategies effectively combat disinformation and ensure that critical messages are understood by all.
Alch3mist, (aka Anthony Nixon) is a web3 engineer with a passion for cognitive science, AI, and information theory. Currently contributing to TEN.
